web-search

Pass

Audited by Gen Agent Trust Hub on Sep 14, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill documentation describes fetching custom ranking configurations (Goggles) from the vendor's official GitHub repository at github.com/brave/goggles-quickstart. This is a vendor-owned resource used for search result filtering.
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted external data from the web, including search snippets, page titles, and forum comments, which are subsequently processed by the agent.
  • Ingestion points: Response fields documented in SKILL.md such as web.results[].description, web.results[].title, and discussions.results[].data.top_comment.
  • Boundary markers: The documentation does not specify the use of delimiters or warnings to ignore embedded instructions within search results.
  • Capability inventory: The skill documentation includes examples of network operations via curl to the api.search.brave.com endpoint.
  • Sanitization: The skill does not mention specific sanitization or filtering of the content retrieved from external websites.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 14, 2026, 05:31 PM
Security Audit — agent-trust-hub — web-search